Latency function estimation under the mixture cure model when the cure status is available.

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Mixture cure models are used when a proportion of subjects may never experience the event of interest.
• Traditional methods assume long-term survivors are unidentifiable due to right censoring.
• This assumption is violated when cure status can be definitively determined (e.g., via medical tests).

Purpose of the Study:

• To develop a novel latency estimator for mixture cure models with partially available cure status information.
• To extend existing nonparametric estimators to accommodate situations with confirmed cured subjects.
• To provide a robust statistical tool for survival analysis in complex scenarios.

Main Methods:

• Proposed a nonparametric latency estimator building upon prior work (López-Cheda et al., 2017).
• The estimator is adapted for scenarios where cure status is partially observed.
• Asymptotic normality distribution of the proposed estimator was established.

Main Results:

• The developed estimator effectively handles partially available cure status information.
• Simulation studies demonstrated the estimator's performance.
• The estimator's applicability was validated on a real-world medical dataset.

Conclusions:

• The proposed latency estimator offers a valuable extension for mixture cure models with partially known cure statuses.
• This method enhances the accuracy of survival function estimation in such cases.
• The approach has practical implications for analyzing patient outcomes, such as COVID-19 intensive care unit length of stay.